Model for Spatiotemporal Crime Prediction with Improved Deep Learning

Crime is hard to anticipate since it occurs at random and can occur anywhere any moment, making a difficult issue for society address. By analyzing comparing eight known prediction models: Naive Bayes, Stacking, Random Forest, Lazy:IBK, Bagging, Support Vector Machine, Convolutional Neural Network, Locally Weighted Learning – this study proposed an improved deep learning crime model using convolutional neural networks the xgboost algorithm predict crime. The major goal of research provide based on previous criminal records. Using Boston dataset, where our larceny dataset was extracted, exploratory data analysis (EDA) used uncover patterns explain trends in crimes. performance basis accuracy, recall, f-measure 100% outperforming other models study. aid security services better use their resources, anticipating certain time, serving better.